Visualization and Analysis of IncuCyte Output
Maia Smith
February 9, 2015
## Warning: package 'car' was built under R version 3.1.3## Loading required package: RCurl
## Loading required package: bitops
## Loading required package: RJSONIO
##
## Attaching package: 'gplots'
##
## The following object is masked from 'package:stats':
##
## lowessQuality Control
Compare controls and treatments
Compare plates for controls and treatments
Compare sparklines for each plate, controls vs treatments, with mean and sd for each plate and control/treatment
Sparklines for Each Compound
Pathway analysis
Sparklines by Pathway
Compared to negative controls:
In Plotly:
Delta (max-min) values by Pathway
Heatmap of metrics for all compounds (not negative controls) using Sytox Green Data
Comparison to Negative Controls
Cluster Analysis
# Parameters
start_index <- which(colnames(confluency_sytoxG_data_prelim_proc) == "0")
end_index <- which(colnames(confluency_sytoxG_data_prelim_proc) == "46")
num_time_intervals <- length(unique(sytoxG_data$time_elapsed)) # Number of time intervals
num_clusters <- 12
# Confidence interval bounds
confidence_intervals_SG <- sytoxG_data[1:num_time_intervals,c("time_elapsed", "phenotype_value.NC.upper", "phenotype_value.NC.mean", "phenotype_value.NC.lower")]
# Get Sytox Green raw time series data only, for all compounds (including negative controls)
data_for_heatmap <- confluency_sytoxG_data_prelim_proc[confluency_sytoxG_data_prelim_proc$phenotypic_Marker == "SG", start_index:end_index]
rownames(data_for_heatmap) <- confluency_sytoxG_data_prelim_proc[confluency_sytoxG_data_prelim_proc$phenotypic_Marker == "SG",]$Compound
data_matrix <- as.matrix(data_for_heatmap)
# Get distances (Euclidean) and clusters
distMatrix <- dist(data_matrix, method="euclidean")
hr <- hclust(distMatrix, method="average")
# Cut the tree and create color vector for clusters.
mycl <- cutree(hr, k = num_clusters) # Clusters assigned to each compound.
mycolhc <- rainbow(length(unique(mycl)), start=0.1, end=0.9)
mycolhc <- mycolhc[as.vector(mycl)]
# Plot all clusters
compound_clusters <- as.data.frame(mycl)
colnames(compound_clusters) <- "cluster"
compound_clusters$cluster <- as.factor(compound_clusters$cluster)
sytoxG_data_w_clusters <- merge(sytoxG_data, compound_clusters, by.x="Compound", by.y="row.names")
ggplot(sytoxG_data_w_clusters) +
geom_line(aes(x=as.numeric(time_elapsed), y=as.numeric(phenotype_value), group=Compound, text=Compound)) +
geom_ribbon(data = confidence_intervals_SG, mapping = aes(x = time_elapsed, ymin = phenotype_value.NC.lower, ymax = phenotype_value.NC.upper,
fill = "red", colour = NULL), alpha = 0.6) +
scale_fill_manual(name = "Legend",
values = c('red'),
labels = c('Negative Control\n99.9% C.I.')) +
xlab("Time Elapsed") +
ylab("Sytox Green") +
ggtitle("Sytox Green Sparklines for Each Cluster") +
facet_grid(empty~cluster) +
theme(panel.grid = element_blank(),
axis.ticks.length = unit(0, "cm"),
panel.background = element_rect(fill = "white"),
strip.text.x = element_text(size=4),
axis.text = element_blank())
Statistical Analyses
Q-Q Plots
## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).
## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).
## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).
## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).## Warning: Removed 87 rows containing missing values (stat_qq).
At what time point do we see the first significant differences in phenotypic marker values?
In Sytox Green?
##
## Call:
## lm(formula = phenotype_value ~ as.factor(time_elapsed), data = data_for_stats,
## subset = phenotypic_Marker == "SG")
##
## Residuals:
## Min 1Q Median 3Q Max
## -314.46 -38.23 -14.59 8.44 2669.95
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 74.528 2.483 30.016 < 2e-16 ***
## as.factor(time_elapsed)2 4.108 3.511 1.170 0.24207
## as.factor(time_elapsed)4 2.834 3.511 0.807 0.41970
## as.factor(time_elapsed)6 1.119 3.511 0.319 0.74997
## as.factor(time_elapsed)8 9.289 3.511 2.645 0.00816 **
## as.factor(time_elapsed)10 30.571 3.511 8.706 < 2e-16 ***
## as.factor(time_elapsed)12 52.287 3.511 14.891 < 2e-16 ***
## as.factor(time_elapsed)14 66.403 3.511 18.911 < 2e-16 ***
## as.factor(time_elapsed)16 75.650 3.511 21.544 < 2e-16 ***
## as.factor(time_elapsed)18 87.923 3.511 25.039 < 2e-16 ***
## as.factor(time_elapsed)20 109.096 3.511 31.069 < 2e-16 ***
## as.factor(time_elapsed)22 138.903 3.511 39.557 < 2e-16 ***
## as.factor(time_elapsed)24 170.426 3.511 48.535 < 2e-16 ***
## as.factor(time_elapsed)26 196.007 3.511 55.820 < 2e-16 ***
## as.factor(time_elapsed)28 214.572 3.511 61.107 < 2e-16 ***
## as.factor(time_elapsed)30 227.747 3.511 64.859 < 2e-16 ***
## as.factor(time_elapsed)32 236.130 3.511 67.246 < 2e-16 ***
## as.factor(time_elapsed)34 240.114 3.511 68.380 < 2e-16 ***
## as.factor(time_elapsed)36 240.164 3.511 68.395 < 2e-16 ***
## as.factor(time_elapsed)38 237.577 3.511 67.658 < 2e-16 ***
## as.factor(time_elapsed)40 233.288 3.511 66.436 < 2e-16 ***
## as.factor(time_elapsed)42 229.036 3.511 65.226 < 2e-16 ***
## as.factor(time_elapsed)44 225.055 3.511 64.092 < 2e-16 ***
## as.factor(time_elapsed)46 221.107 3.511 62.968 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 108.8 on 46056 degrees of freedom
## Multiple R-squared: 0.4277, Adjusted R-squared: 0.4274
## F-statistic: 1497 on 23 and 46056 DF, p-value: < 2.2e-16## [1] "The first significant time point for Sytox Green after the hour 0 is hour 8."In Confluency?
##
## Call:
## lm(formula = phenotype_value ~ as.factor(time_elapsed), data = data_for_stats,
## subset = phenotypic_Marker == "Con")
##
## Residuals:
## Min 1Q Median 3Q Max
## -27.9372 -0.2166 0.6207 1.2699 2.3732
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 99.20075 0.05590 1774.520 < 2e-16 ***
## as.factor(time_elapsed)2 0.08931 0.07906 1.130 0.258628
## as.factor(time_elapsed)4 0.02492 0.07906 0.315 0.752606
## as.factor(time_elapsed)6 -0.10445 0.07906 -1.321 0.186459
## as.factor(time_elapsed)8 -0.27530 0.07906 -3.482 0.000498 ***
## as.factor(time_elapsed)10 -0.43711 0.07906 -5.529 3.24e-08 ***
## as.factor(time_elapsed)12 -0.59151 0.07906 -7.482 7.46e-14 ***
## as.factor(time_elapsed)14 -0.67342 0.07906 -8.518 < 2e-16 ***
## as.factor(time_elapsed)16 -0.73024 0.07906 -9.237 < 2e-16 ***
## as.factor(time_elapsed)18 -0.70453 0.07906 -8.911 < 2e-16 ***
## as.factor(time_elapsed)20 -0.68427 0.07906 -8.655 < 2e-16 ***
## as.factor(time_elapsed)22 -0.65260 0.07906 -8.255 < 2e-16 ***
## as.factor(time_elapsed)24 -0.63643 0.07906 -8.050 8.47e-16 ***
## as.factor(time_elapsed)26 -0.60346 0.07906 -7.633 2.34e-14 ***
## as.factor(time_elapsed)28 -0.61336 0.07906 -7.758 8.78e-15 ***
## as.factor(time_elapsed)30 -0.63689 0.07906 -8.056 8.08e-16 ***
## as.factor(time_elapsed)32 -0.68883 0.07906 -8.713 < 2e-16 ***
## as.factor(time_elapsed)34 -0.76726 0.07906 -9.705 < 2e-16 ***
## as.factor(time_elapsed)36 -0.83190 0.07906 -10.523 < 2e-16 ***
## as.factor(time_elapsed)38 -0.97516 0.07906 -12.335 < 2e-16 ***
## as.factor(time_elapsed)40 -1.11853 0.07906 -14.148 < 2e-16 ***
## as.factor(time_elapsed)42 -1.27866 0.07906 -16.174 < 2e-16 ***
## as.factor(time_elapsed)44 -1.42124 0.07906 -17.977 < 2e-16 ***
## as.factor(time_elapsed)46 -1.57391 0.07906 -19.908 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.45 on 46056 degrees of freedom
## Multiple R-squared: 0.02796, Adjusted R-squared: 0.02748
## F-statistic: 57.6 on 23 and 46056 DF, p-value: < 2.2e-16## [1] "The first significant time point for Confluency after the hour 0 is hour 8."Which pathways are significantly different from the negative controls (measured by AUC)?
For Sytox Green:
## [1] "The following pathways have an AUC for Sytox Green significantly different (p < 0.01) from the negative controls' AUC?"## [1] "Apoptosis" "Cell Cycle"
## [3] "DNA Damage" "Endocrinology & Hormones"
## [5] "GPCR & G Protein" "JAK/STAT"
## [7] "MAPK" "Metabolism"
## [9] "Microbiology" "NegControl"
## [11] "Neuronal Signaling" "NF-B"
## [13] "Others" "PI3K/Akt/mTOR"
## [15] "Proteases" "Protein Tyrosine Kinase"
## [17] "Stem Cells & Wnt" "TGF-beta/Smad"
## [19] "Transmembrane Transporters" "Ubiquitin"For Confluency:
## [1] "The following pathways have an AUC for Confluency significantly different (p < 0.01) from the negative controls' AUC?"## [1] "Apoptosis" "Cell Cycle"
## [3] "Endocrinology & Hormones" "Epigenetics"
## [5] "GPCR & G Protein" "JAK/STAT"
## [7] "MAPK" "NegControl"
## [9] "NF-B" "Others"
## [11] "PI3K/Akt/mTOR" "Proteases"
## [13] "Stem Cells & Wnt" "Ubiquitin"Which targets are significantly different from the negative controls (measured by AUC)?
For Sytox Green:
## [1] "The following targets have an AUC for Sytox Green significantly different (p < 1e-20) from the negative controls' AUC?"## [1] "Akt,mTOR,PI3K" "ATM/ATR,mTOR"
## [3] "Autophagy,Bcl-2" "Bcl-2,Autophagy"
## [5] "Bcr-Abl,c-Kit,Src" "c-Met"
## [7] "c-RET,FGFR,Bcr-Abl,Aurora Kinase" "CDK"
## [9] "Chk" "DNA-PK,PDGFR,mTOR"
## [11] "DUB,Bcr-Abl" "EGFR"
## [13] "Epigenetic Reader Domain" "HDAC"
## [15] "HDAC,HER2,EGFR" "HER2,VEGFR,EGFR"
## [17] "HSP (e.g. HSP90),Autophagy" "IB/IKK"
## [19] "IB/IKK,E2 conjugating" "JAK,FLT3,c-RET"
## [21] "JNK" "mTOR,PI3K"
## [23] "PDGFR,FGFR,VEGFR,Bcr-Abl" "PI3K,Autophagy,DNA-PK,mTOR"
## [25] "PI3K,DNA-PK" "PI3K,HDAC"
## [27] "PI3K,mTOR" "PKA,CDK,Akt"
## [29] "PKC" "PLK"
## [31] "Proton Pump" "STAT"
## [33] "Survivin" "Syk"
## [35] "Topoisomerase" "VEGFR,PDGFR,c-Kit"Sytox Green sparklines for the significant targets:
For Confluency:
## [1] "The following targets have an AUC for Confluency significantly different (p < 1e-20) from the negative controls' AUC?"## [1] "ATPase,Autophagy"
## [2] "Aurora Kinase,FLT3,VEGFR"
## [3] "Autophagy,Bcl-2"
## [4] "c-Kit,VEGFR,PDGFR"
## [5] "CaSR"
## [6] "CRM1"
## [7] "DNA-PK,PI3K"
## [8] "DUB"
## [9] "FLT3,Tie-2,c-Kit,c-Met,VEGFR,Axl"
## [10] "GSK-3"
## [11] "IB/IKK"
## [12] "IB/IKK,E2 conjugating"
## [13] "IDO"
## [14] "p53"
## [15] "p97"
## [16] "PAK"
## [17] "Proteasome"
## [18] "Proton Pump"
## [19] "Raf,Src,Bcr-Abl,VEGFR,Ephrin receptor"
## [20] "S1P Receptor"
## [21] "STAT"
## [22] "TNF-alpha,NF-B"
## [23] "Topoisomerase"Confluency sparklines for the significant targets:
